rtnetlink: fdb dump: optimize by saving last interface markers

fdb dumps spanning multiple skb's currently restart from the first
interface again for every skb. This results in unnecessary
iterations on the already visited interfaces and their fdb
entries. In large scale setups, we have seen this to slow
down fdb dumps considerably. On a system with 30k macs we
see fdb dumps spanning across more than 300 skbs.

To fix the problem, this patch replaces the existing single fdb
marker with three markers: netdev hash entries, netdevs and fdb
index to continue where we left off instead of restarting from the
first netdev. This is consistent with link dumps.

In the process of fixing the performance issue, this patch also
re-implements fix done by
commit 472681d57a5d ("net: ndo_fdb_dump should report -EMSGSIZE to rtnl_fdb_dump")
(with an internal fix from Wilson Kok) in the following ways:
- change ndo_fdb_dump handlers to return error code instead
of the last fdb index
- use cb->args strictly for dump frag markers and not error codes.
This is consistent with other dump functions.

Below results were taken on a system with 1000 netdevs
and 35085 fdb entries:
before patch:
$time bridge fdb show | wc -l
15065

real    1m11.791s
user    0m0.070s
sys 1m8.395s

(existing code does not return all macs)

after patch:
$time bridge fdb show | wc -l
35085

real    0m2.017s
user    0m0.113s
sys 0m1.942s

Signed-off-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: Wilson Kok <wkok@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 65 insertions, 40 deletions

diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
index 318fc5231b2b..1dfca1c3f8f5 100644
--- a/net/core/rtnetlink.c
+++ b/net/core/rtnetlink.c
@@ -3068,7 +3068,7 @@ static int nlmsg_populate_fdb(struct sk_buff *skb,
 	seq = cb->nlh->nlmsg_seq;
 
 	list_for_each_entry(ha, &list->list, list) {
-		if (*idx < cb->args[0])
+		if (*idx < cb->args[2])
 			goto skip;
 
 		err = nlmsg_populate_fdb_fill(skb, dev, ha->addr, 0,
@@ -3095,19 +3095,18 @@ int ndo_dflt_fdb_dump(struct sk_buff *skb,
 		      struct netlink_callback *cb,
 		      struct net_device *dev,
 		      struct net_device *filter_dev,
-		      int idx)
+		      int *idx)
 {
 	int err;
 
 	netif_addr_lock_bh(dev);
-	err = nlmsg_populate_fdb(skb, cb, dev, &idx, &dev->uc);
+	err = nlmsg_populate_fdb(skb, cb, dev, idx, &dev->uc);
 	if (err)
 		goto out;
-	nlmsg_populate_fdb(skb, cb, dev, &idx, &dev->mc);
+	nlmsg_populate_fdb(skb, cb, dev, idx, &dev->mc);
 out:
 	netif_addr_unlock_bh(dev);
-	cb->args[1] = err;
-	return idx;
+	return err;
 }
 EXPORT_SYMBOL(ndo_dflt_fdb_dump);
 
@@ -3120,9 +3119,13 @@ static int rtnl_fdb_dump(struct sk_buff *skb, struct netlink_callback *cb)
 	const struct net_device_ops *cops = NULL;
 	struct ifinfomsg *ifm = nlmsg_data(cb->nlh);
 	struct net *net = sock_net(skb->sk);
+	struct hlist_head *head;
 	int brport_idx = 0;
 	int br_idx = 0;
-	int idx = 0;
+	int h, s_h;
+	int idx = 0, s_idx;
+	int err = 0;
+	int fidx = 0;
 
 	if (nlmsg_parse(cb->nlh, sizeof(struct ifinfomsg), tb, IFLA_MAX,
 			ifla_policy) == 0) {
@@ -3140,49 +3143,71 @@ static int rtnl_fdb_dump(struct sk_buff *skb, struct netlink_callback *cb)
 		ops = br_dev->netdev_ops;
 	}
 
-	cb->args[1] = 0;
-	for_each_netdev(net, dev) {
-		if (brport_idx && (dev->ifindex != brport_idx))
-			continue;
+	s_h = cb->args[0];
+	s_idx = cb->args[1];
 
-		if (!br_idx) { /* user did not specify a specific bridge */
-			if (dev->priv_flags & IFF_BRIDGE_PORT) {
-				br_dev = netdev_master_upper_dev_get(dev);
-				cops = br_dev->netdev_ops;
-			}
+	for (h = s_h; h < NETDEV_HASHENTRIES; h++, s_idx = 0) {
+		idx = 0;
+		head = &net->dev_index_head[h];
+		hlist_for_each_entry(dev, head, index_hlist) {
 
-		} else {
-			if (dev != br_dev &&
-			    !(dev->priv_flags & IFF_BRIDGE_PORT))
+			if (brport_idx && (dev->ifindex != brport_idx))
 				continue;
 
-			if (br_dev != netdev_master_upper_dev_get(dev) &&
-			    !(dev->priv_flags & IFF_EBRIDGE))
-				continue;
+			if (!br_idx) { /* user did not specify a specific bridge */
+				if (dev->priv_flags & IFF_BRIDGE_PORT) {
+					br_dev = netdev_master_upper_dev_get(dev);
+					cops = br_dev->netdev_ops;
+				}
+			} else {
+				if (dev != br_dev &&
+				    !(dev->priv_flags & IFF_BRIDGE_PORT))
+					continue;
 
-			cops = ops;
-		}
+				if (br_dev != netdev_master_upper_dev_get(dev) &&
+				    !(dev->priv_flags & IFF_EBRIDGE))
+					continue;
+				cops = ops;
+			}
 
-		if (dev->priv_flags & IFF_BRIDGE_PORT) {
-			if (cops && cops->ndo_fdb_dump)
-				idx = cops->ndo_fdb_dump(skb, cb, br_dev, dev,
-							 idx);
-		}
-		if (cb->args[1] == -EMSGSIZE)
-			break;
+			if (idx < s_idx)
+				goto cont;
 
-		if (dev->netdev_ops->ndo_fdb_dump)
-			idx = dev->netdev_ops->ndo_fdb_dump(skb, cb, dev, NULL,
-							    idx);
-		else
-			idx = ndo_dflt_fdb_dump(skb, cb, dev, NULL, idx);
-		if (cb->args[1] == -EMSGSIZE)
-			break;
+			if (dev->priv_flags & IFF_BRIDGE_PORT) {
+				if (cops && cops->ndo_fdb_dump) {
+					err = cops->ndo_fdb_dump(skb, cb,
+								br_dev, dev,
+								&fidx);
+					if (err == -EMSGSIZE)
+						goto out;
+				}
+			}
 
-		cops = NULL;
+			if (dev->netdev_ops->ndo_fdb_dump)
+				err = dev->netdev_ops->ndo_fdb_dump(skb, cb,
+								    dev, NULL,
+								    &fidx);
+			else
+				err = ndo_dflt_fdb_dump(skb, cb, dev, NULL,
+							&fidx);
+			if (err == -EMSGSIZE)
+				goto out;
+
+			cops = NULL;
+
+			/* reset fdb offset to 0 for rest of the interfaces */
+			cb->args[2] = 0;
+			fidx = 0;
+cont:
+			idx++;
+		}
 	}
 
-	cb->args[0] = idx;
+out:
+	cb->args[0] = h;
+	cb->args[1] = idx;
+	cb->args[2] = fidx;
+
 	return skb->len;
 }